An isobaric polynomial is a symmetric polynomial written on the Elementary Symmetric Polynomial b... more An isobaric polynomial is a symmetric polynomial written on the Elementary Symmetric Polynomial basis [1]. These polynomials have the pleasant property of being indexed by partitions of the natural numbers, and they turn out to provide useful representations of several arithmetic and algebraic structures, [4], [5].. We introduce two new operators, the Isobaric logarithm and the isobaric exponential operator from which we can define the hyperbolic trigonometric functions: the isobaric hyperbolic sine, cosine, tangent, etc. for which all of the usual hyperbolic trigonometric identities are satisfied. Since the ring of isobaric polynomials is isomorphic to the ring of symmetric polynomials, all of these identities induce (new) identities in the ring of symmetric polynomials, and imply the existence of a hyperbolic geometric structure for the ring of symmetric polynomials. We are dealing with an algebra (or rather two isomorphic algebras), which has the property of supporting a hyperbol...
Dental caries is an extremely common problem in dentistry that affects a significant part of the ... more Dental caries is an extremely common problem in dentistry that affects a significant part of the population. Approximal caries are especially difficult to identify because their position makes clinical analysis difficult. Radiographic evaluation—more specifically, bitewing images—are mostly used in such cases. However, incorrect interpretations may interfere with the diagnostic process. To aid dentists in caries evaluation, computational methods and tools can be used. In this work, we propose a new method that combines image processing techniques and convolutional neural networks to identify approximal dental caries in bitewing radiographic images and classify them according to lesion severity. For this study, we acquired 112 bitewing radiographs. From these exams, we extracted individual tooth images from each exam, applied a data augmentation process, and used the resulting images to train CNN classification models. The tooth images were previously labeled by experts to denote the...
Anais do XXI Simpósio Brasileiro de Computação Aplicada à Saúde (SBCAS 2021)
Radiografias são ferramentas indispensáveis para auxílio ao diagnóstico médico. Ainda que outras t... more Radiografias são ferramentas indispensáveis para auxílio ao diagnóstico médico. Ainda que outras técnicas estejam disponíveis, a radiografia é um exame acessível, de rápida aquisição e utilizado em larga escala. Este trabalho apresenta um método de segmentação automatizada das áreas pulmonares em radiografias torácicas e para tal, faz uso de técnicas de processamento de imagens simples. Disponibilizamos publicamente as segmentações desenvolvidas; resultados são comparados aos de outras técnicas disponíveis na literatura e podem ser utilizados como entrada em sistemas de auxílio ao diagnóstico.
Anais do XXI Simpósio Brasileiro de Computação Aplicada à Saúde (SBCAS 2021)
O câncer de mama é o principal tipo de câncer entre as mulheres. De acordo com o World Cancer Res... more O câncer de mama é o principal tipo de câncer entre as mulheres. De acordo com o World Cancer Research Fund, em 2018, mais de 2 milhões de novos casos foram detectados em todo o mundo. Apesar de sua alta ocorrência, a detecção precoce proporciona um melhor prognóstico e auxilia no aumento da sobrevida do paciente oncológico. Avanços significativos nas técnicas de rastreamento, como as imagens infravermelhas, forneceram uma maneira barata e menos invasiva forma de detectar a doença. Além disso, ferramentas computacionais podem ser utilizadas para auxiliar os médicos a fornecerem um melhor diagnóstico. Assim, este artigo apresenta um método de segmentação baseado em Redes Neurais Convolucionais U-Net. Em contraste com o estado da arte, as abordagens de aprendizado de máquina têm se mostrado eficientes para a segmentação da região de interesse deste trabalho, atingindo uma acurácia de 98,24% e uma Intersecção-Sobre-União de 94,38%. O uso deste método de segmentação pode ser muito útil pa...
Convolutional Neural Networks (CNNs) have been successfully applied in the medical diagnosis of d... more Convolutional Neural Networks (CNNs) have been successfully applied in the medical diagnosis of different types of diseases. However, selecting the architecture and the best set of hyperparameters among the possible combinations can be a significant challenge. The purpose of this work is to investigate the use of the Hyperband optimization algorithm in the process of optimizing a CNN applied to the diagnosis of SARS-Cov2 disease (COVID-19). The test was performed with the Optuna framework, and the optimization process aimed to optimize four hyperparameters: (1) backbone architecture, (2) the number of inception modules, (3) the number of neurons in the fully connected layers, and (4) the learning rate. CNNs were trained on 2175 computed tomography (CT) images. The CNN that was proposed by the optimization process was a VGG16 with five inception modules, 128 neurons in the two fully connected layers, and a learning rate of 0.0027. The proposed method achieved a sensitivity, precision...
Resolution plays an essential role in oral imaging for periodontal disease assessment. Neverthele... more Resolution plays an essential role in oral imaging for periodontal disease assessment. Nevertheless, due to limitations in acquisition tools, a considerable number of oral examinations have low resolution, making the evaluation of this kind of lesion difficult. Recently, the use of deep-learning methods for image resolution improvement has seen an increase in the literature. In this work, we performed two studies to evaluate the effects of using different resolution improvement methods (nearest, bilinear, bicubic, Lanczos, SRCNN, and SRGAN). In the first one, specialized dentists visually analyzed the quality of images treated with these techniques. In the second study, we used those methods as different pre-processing steps for inputs of convolutional neural network (CNN) classifiers (Inception and ResNet) and evaluated whether this process leads to better results. The deep-learning methods lead to a substantial improvement in the visual quality of images but do not necessarily pro...
Breast cancer has been the second leading cause of cancer death among women. New techniques to en... more Breast cancer has been the second leading cause of cancer death among women. New techniques to enhance early diagnosis are very important to improve cure rates. This paper proposes and evaluates an image analysis method to automatically detect patients with breast benign and malignant changes (tumors). Such method explores the difference of Dynamic Infrared Thermography (DIT) patterns observed in patients’ skin. After obtaining the sequential DIT images of each patient, their temperature arrays are computed and new images in gray scale are generated. Then the regions of interest (ROIs) of those images are segmented and, from them, arrays of the ROI temperature are computed. Features are extracted from the arrays, such as the ones based on statistical, clustering, histogram comparison, fractal geometry, diversity indices and spatial statistics. Time series that are broken down into subsets of different cardinalities are generated from such features. Automatic feature selection method...
According to experts and medical literature, a healthy thyroid gland, or a thyroid containing ben... more According to experts and medical literature, a healthy thyroid gland, or a thyroid containing benign nodules, tend to be less inflamed and less active than one with malignant nodules. It seems to be a consensus that malignant nodules have more blood veins and it may be related to the maintenance of high and constant temperatures. Investigation of these characteristics, detectable by infrared sensors, and answering if they constitute patterns of malignancy are the aims of this work. Experiments considering biological heat transfer analysis by Finite Element numerical simulations are used to show the influence of nodule and patient characteristics on the identification of malignancy of thyroid nodule by thermography. The used and approved protocol for infrared examination are analyzed and simulated during all its phase that is on transient and steady state behavior, in order to verify how and when their influence can be really recognized in patients. Simulation results and the analysi...
Anais Pr ncipais do Simpósio Brasileiro de Computação Aplicada à Saúde (SBCAS 2015)
A reconstrução 3D, aliada à termografia, é um meio de validar, ou ajudar o diagnóstico, tornando-... more A reconstrução 3D, aliada à termografia, é um meio de validar, ou ajudar o diagnóstico, tornando-o mais preciso. Neste artigo é proposta uma metodologia para reconstrução da geometria 3D da mama visando a realização de simulações computacionais, assim como auxiliar no planejamento de cirurgias. A abordagem consiste em três etapas: calibração de dois Kinects; registro das nuvens de pontos adquiridas; reconstrução da superfície do objeto virtual. Nas validações a média da diferença entre as áreas de superfície foi de 3,55%, 0,93 de Coeficiente de similaridade Dice, na média, e a média da diferença das distâncias entre os mamilos foi de 3,51%.
Advances in Science, Technology and Engineering Systems Journal
Breast cancer has an important incidence in women mortality worldwide. Currently, mammography is ... more Breast cancer has an important incidence in women mortality worldwide. Currently, mammography is considered the gold standard for breast abnormalities screening examinations, since it aids in the early detection and diagnosis of the illness. However, both identification of mass lesions and its malignancy classification is a challenging problem for artificial intelligence. In this work, we extend our previous research in mammogram classification, where we studied NasNet and MobileNet in transfer learning to train a breast abnormality malignancy classifier, and include models like: VGG, Resnet, Xception and Resnext. However, training deep learning models tends to overfit. This problem is also carried out in this work. Our results show that Fine Tuning achieves the best classifier performance in VGG16 with AUC value of 0.844 in the CBIS-DDSM dataset.
Anais do Simpósio Brasileiro de Computação Aplicada à Saúde (SBCAS)
Medical images usually must have their region of interest (ROI) segmented as a first step in a pa... more Medical images usually must have their region of interest (ROI) segmented as a first step in a pattern recognition procedure. Automatic segmentation of these images is an open issue. This paper presents an automated technique to define the ROI for infrared breast exams, based on the use of Fully Convolutional Networks (FCN). Adequate comparison among new approaches by using available databases is very important, here some comparisons with other techniques are made. Moreover, concerning on line diagnosis, the comparison among possible techniques must be efficient enough to be done in real time. With our approach the time to segment the ROI was 100 milliseconds and the average accuracy obtained was 95%.
Anais do Simpósio Brasileiro de Sistemas de Informação (SBSI)
In a scenario where there is a huge amount of available data sources, the Semantic Web has played... more In a scenario where there is a huge amount of available data sources, the Semantic Web has played a key role in sharing, retrieval, selection, and combination of data organized in various formats. The storage and retrieval of medical images manipulated by systems that support breast cancer detection can take great advantage from the use of such technology. In this paper we present a comprehensive study on ontology-based systems that support the manipulation of medical images related to breast cancer, identifying the main features of each approach.
THE 9TH INTERNATIONAL CONFERENCE ON STRUCTURAL ANALYSIS OF ADVANCED MATERIALS - ICSAAM 2019
Estimation of the stresses acting on the borehole wall is a very important aspect in oil industry... more Estimation of the stresses acting on the borehole wall is a very important aspect in oil industry. This work presents an image processing based tool for analyzing the well safety from borehole imaging instruments, because despite the advances on the development of acquisition tools, the final interpretation remains heavily dependent on the skill, experience and alertness of a human. Existing computational tools for the most used equipment fail to detect all fracture types and do not characterize a number of important occurrences on the borehole. This work presents an approach to help in the characterization of damages from ultrasound data. A number of techniques are combined with the amplitude and transit time data available to enable distinctions between fractures, stratifications, axial displacements, porous and breakouts.
Este trabalho propõe uma metodologia para identificar regiões suspeitas de lesão baseada nas assi... more Este trabalho propõe uma metodologia para identificar regiões suspeitas de lesão baseada nas assime-trias da mama esquerda e direita de imagens de termogramas. O estudo é pautado em imagens de pacientes do Hospital Universitário da Universidade Federal de Pernambuco (UFPE), capturadas por câmera infravermelha. Inicialmente as imagens são manualmente segmentadas. Em seguida, os seios são registrados usando a transformação B-spline. Além disso, como o corpo humano tem uma simetria radial das temperaturas, uma lesão, eventualmente, leva uma assimetria destas regiões, em seguida, o spatiogram é usado para identificar essas regiões assimétricas. Finalmente, apenas as regiões com temperaturas superiores à média são mantidas, com base no fato de que o câncer tem a temperatura mais elevada do que o restante mama. Após esse processo são extraídas características (Variação dos pixels, a média, o desvio padrão, o índice de Geary e Dimensão Fractal de Higuchi) para a classificação dessas regiõe...
Tumor growth is a complex process that requires mathematical modeling approaches for studying rea... more Tumor growth is a complex process that requires mathematical modeling approaches for studying real-life cancer behavior. The use of cellular automata (CA) to represent tumor growth in its avascular stage is explained in this work, and a stochastic CA describing tumor growth is obtained, based on a differential equations system in the range of continuum mechanics. The novelty of this research is the deduction of the neighborhood structure and rules for a probabilistic CA from these differential equations that describe the evolution of the tumor growth. In addition, the influence of the stresses on tumor growth is captured by the CA.
An isobaric polynomial is a symmetric polynomial written on the Elementary Symmetric Polynomial b... more An isobaric polynomial is a symmetric polynomial written on the Elementary Symmetric Polynomial basis [1]. These polynomials have the pleasant property of being indexed by partitions of the natural numbers, and they turn out to provide useful representations of several arithmetic and algebraic structures, [4], [5].. We introduce two new operators, the Isobaric logarithm and the isobaric exponential operator from which we can define the hyperbolic trigonometric functions: the isobaric hyperbolic sine, cosine, tangent, etc. for which all of the usual hyperbolic trigonometric identities are satisfied. Since the ring of isobaric polynomials is isomorphic to the ring of symmetric polynomials, all of these identities induce (new) identities in the ring of symmetric polynomials, and imply the existence of a hyperbolic geometric structure for the ring of symmetric polynomials. We are dealing with an algebra (or rather two isomorphic algebras), which has the property of supporting a hyperbol...
Dental caries is an extremely common problem in dentistry that affects a significant part of the ... more Dental caries is an extremely common problem in dentistry that affects a significant part of the population. Approximal caries are especially difficult to identify because their position makes clinical analysis difficult. Radiographic evaluation—more specifically, bitewing images—are mostly used in such cases. However, incorrect interpretations may interfere with the diagnostic process. To aid dentists in caries evaluation, computational methods and tools can be used. In this work, we propose a new method that combines image processing techniques and convolutional neural networks to identify approximal dental caries in bitewing radiographic images and classify them according to lesion severity. For this study, we acquired 112 bitewing radiographs. From these exams, we extracted individual tooth images from each exam, applied a data augmentation process, and used the resulting images to train CNN classification models. The tooth images were previously labeled by experts to denote the...
Anais do XXI Simpósio Brasileiro de Computação Aplicada à Saúde (SBCAS 2021)
Radiografias são ferramentas indispensáveis para auxílio ao diagnóstico médico. Ainda que outras t... more Radiografias são ferramentas indispensáveis para auxílio ao diagnóstico médico. Ainda que outras técnicas estejam disponíveis, a radiografia é um exame acessível, de rápida aquisição e utilizado em larga escala. Este trabalho apresenta um método de segmentação automatizada das áreas pulmonares em radiografias torácicas e para tal, faz uso de técnicas de processamento de imagens simples. Disponibilizamos publicamente as segmentações desenvolvidas; resultados são comparados aos de outras técnicas disponíveis na literatura e podem ser utilizados como entrada em sistemas de auxílio ao diagnóstico.
Anais do XXI Simpósio Brasileiro de Computação Aplicada à Saúde (SBCAS 2021)
O câncer de mama é o principal tipo de câncer entre as mulheres. De acordo com o World Cancer Res... more O câncer de mama é o principal tipo de câncer entre as mulheres. De acordo com o World Cancer Research Fund, em 2018, mais de 2 milhões de novos casos foram detectados em todo o mundo. Apesar de sua alta ocorrência, a detecção precoce proporciona um melhor prognóstico e auxilia no aumento da sobrevida do paciente oncológico. Avanços significativos nas técnicas de rastreamento, como as imagens infravermelhas, forneceram uma maneira barata e menos invasiva forma de detectar a doença. Além disso, ferramentas computacionais podem ser utilizadas para auxiliar os médicos a fornecerem um melhor diagnóstico. Assim, este artigo apresenta um método de segmentação baseado em Redes Neurais Convolucionais U-Net. Em contraste com o estado da arte, as abordagens de aprendizado de máquina têm se mostrado eficientes para a segmentação da região de interesse deste trabalho, atingindo uma acurácia de 98,24% e uma Intersecção-Sobre-União de 94,38%. O uso deste método de segmentação pode ser muito útil pa...
Convolutional Neural Networks (CNNs) have been successfully applied in the medical diagnosis of d... more Convolutional Neural Networks (CNNs) have been successfully applied in the medical diagnosis of different types of diseases. However, selecting the architecture and the best set of hyperparameters among the possible combinations can be a significant challenge. The purpose of this work is to investigate the use of the Hyperband optimization algorithm in the process of optimizing a CNN applied to the diagnosis of SARS-Cov2 disease (COVID-19). The test was performed with the Optuna framework, and the optimization process aimed to optimize four hyperparameters: (1) backbone architecture, (2) the number of inception modules, (3) the number of neurons in the fully connected layers, and (4) the learning rate. CNNs were trained on 2175 computed tomography (CT) images. The CNN that was proposed by the optimization process was a VGG16 with five inception modules, 128 neurons in the two fully connected layers, and a learning rate of 0.0027. The proposed method achieved a sensitivity, precision...
Resolution plays an essential role in oral imaging for periodontal disease assessment. Neverthele... more Resolution plays an essential role in oral imaging for periodontal disease assessment. Nevertheless, due to limitations in acquisition tools, a considerable number of oral examinations have low resolution, making the evaluation of this kind of lesion difficult. Recently, the use of deep-learning methods for image resolution improvement has seen an increase in the literature. In this work, we performed two studies to evaluate the effects of using different resolution improvement methods (nearest, bilinear, bicubic, Lanczos, SRCNN, and SRGAN). In the first one, specialized dentists visually analyzed the quality of images treated with these techniques. In the second study, we used those methods as different pre-processing steps for inputs of convolutional neural network (CNN) classifiers (Inception and ResNet) and evaluated whether this process leads to better results. The deep-learning methods lead to a substantial improvement in the visual quality of images but do not necessarily pro...
Breast cancer has been the second leading cause of cancer death among women. New techniques to en... more Breast cancer has been the second leading cause of cancer death among women. New techniques to enhance early diagnosis are very important to improve cure rates. This paper proposes and evaluates an image analysis method to automatically detect patients with breast benign and malignant changes (tumors). Such method explores the difference of Dynamic Infrared Thermography (DIT) patterns observed in patients’ skin. After obtaining the sequential DIT images of each patient, their temperature arrays are computed and new images in gray scale are generated. Then the regions of interest (ROIs) of those images are segmented and, from them, arrays of the ROI temperature are computed. Features are extracted from the arrays, such as the ones based on statistical, clustering, histogram comparison, fractal geometry, diversity indices and spatial statistics. Time series that are broken down into subsets of different cardinalities are generated from such features. Automatic feature selection method...
According to experts and medical literature, a healthy thyroid gland, or a thyroid containing ben... more According to experts and medical literature, a healthy thyroid gland, or a thyroid containing benign nodules, tend to be less inflamed and less active than one with malignant nodules. It seems to be a consensus that malignant nodules have more blood veins and it may be related to the maintenance of high and constant temperatures. Investigation of these characteristics, detectable by infrared sensors, and answering if they constitute patterns of malignancy are the aims of this work. Experiments considering biological heat transfer analysis by Finite Element numerical simulations are used to show the influence of nodule and patient characteristics on the identification of malignancy of thyroid nodule by thermography. The used and approved protocol for infrared examination are analyzed and simulated during all its phase that is on transient and steady state behavior, in order to verify how and when their influence can be really recognized in patients. Simulation results and the analysi...
Anais Pr ncipais do Simpósio Brasileiro de Computação Aplicada à Saúde (SBCAS 2015)
A reconstrução 3D, aliada à termografia, é um meio de validar, ou ajudar o diagnóstico, tornando-... more A reconstrução 3D, aliada à termografia, é um meio de validar, ou ajudar o diagnóstico, tornando-o mais preciso. Neste artigo é proposta uma metodologia para reconstrução da geometria 3D da mama visando a realização de simulações computacionais, assim como auxiliar no planejamento de cirurgias. A abordagem consiste em três etapas: calibração de dois Kinects; registro das nuvens de pontos adquiridas; reconstrução da superfície do objeto virtual. Nas validações a média da diferença entre as áreas de superfície foi de 3,55%, 0,93 de Coeficiente de similaridade Dice, na média, e a média da diferença das distâncias entre os mamilos foi de 3,51%.
Advances in Science, Technology and Engineering Systems Journal
Breast cancer has an important incidence in women mortality worldwide. Currently, mammography is ... more Breast cancer has an important incidence in women mortality worldwide. Currently, mammography is considered the gold standard for breast abnormalities screening examinations, since it aids in the early detection and diagnosis of the illness. However, both identification of mass lesions and its malignancy classification is a challenging problem for artificial intelligence. In this work, we extend our previous research in mammogram classification, where we studied NasNet and MobileNet in transfer learning to train a breast abnormality malignancy classifier, and include models like: VGG, Resnet, Xception and Resnext. However, training deep learning models tends to overfit. This problem is also carried out in this work. Our results show that Fine Tuning achieves the best classifier performance in VGG16 with AUC value of 0.844 in the CBIS-DDSM dataset.
Anais do Simpósio Brasileiro de Computação Aplicada à Saúde (SBCAS)
Medical images usually must have their region of interest (ROI) segmented as a first step in a pa... more Medical images usually must have their region of interest (ROI) segmented as a first step in a pattern recognition procedure. Automatic segmentation of these images is an open issue. This paper presents an automated technique to define the ROI for infrared breast exams, based on the use of Fully Convolutional Networks (FCN). Adequate comparison among new approaches by using available databases is very important, here some comparisons with other techniques are made. Moreover, concerning on line diagnosis, the comparison among possible techniques must be efficient enough to be done in real time. With our approach the time to segment the ROI was 100 milliseconds and the average accuracy obtained was 95%.
Anais do Simpósio Brasileiro de Sistemas de Informação (SBSI)
In a scenario where there is a huge amount of available data sources, the Semantic Web has played... more In a scenario where there is a huge amount of available data sources, the Semantic Web has played a key role in sharing, retrieval, selection, and combination of data organized in various formats. The storage and retrieval of medical images manipulated by systems that support breast cancer detection can take great advantage from the use of such technology. In this paper we present a comprehensive study on ontology-based systems that support the manipulation of medical images related to breast cancer, identifying the main features of each approach.
THE 9TH INTERNATIONAL CONFERENCE ON STRUCTURAL ANALYSIS OF ADVANCED MATERIALS - ICSAAM 2019
Estimation of the stresses acting on the borehole wall is a very important aspect in oil industry... more Estimation of the stresses acting on the borehole wall is a very important aspect in oil industry. This work presents an image processing based tool for analyzing the well safety from borehole imaging instruments, because despite the advances on the development of acquisition tools, the final interpretation remains heavily dependent on the skill, experience and alertness of a human. Existing computational tools for the most used equipment fail to detect all fracture types and do not characterize a number of important occurrences on the borehole. This work presents an approach to help in the characterization of damages from ultrasound data. A number of techniques are combined with the amplitude and transit time data available to enable distinctions between fractures, stratifications, axial displacements, porous and breakouts.
Este trabalho propõe uma metodologia para identificar regiões suspeitas de lesão baseada nas assi... more Este trabalho propõe uma metodologia para identificar regiões suspeitas de lesão baseada nas assime-trias da mama esquerda e direita de imagens de termogramas. O estudo é pautado em imagens de pacientes do Hospital Universitário da Universidade Federal de Pernambuco (UFPE), capturadas por câmera infravermelha. Inicialmente as imagens são manualmente segmentadas. Em seguida, os seios são registrados usando a transformação B-spline. Além disso, como o corpo humano tem uma simetria radial das temperaturas, uma lesão, eventualmente, leva uma assimetria destas regiões, em seguida, o spatiogram é usado para identificar essas regiões assimétricas. Finalmente, apenas as regiões com temperaturas superiores à média são mantidas, com base no fato de que o câncer tem a temperatura mais elevada do que o restante mama. Após esse processo são extraídas características (Variação dos pixels, a média, o desvio padrão, o índice de Geary e Dimensão Fractal de Higuchi) para a classificação dessas regiõe...
Tumor growth is a complex process that requires mathematical modeling approaches for studying rea... more Tumor growth is a complex process that requires mathematical modeling approaches for studying real-life cancer behavior. The use of cellular automata (CA) to represent tumor growth in its avascular stage is explained in this work, and a stochastic CA describing tumor growth is obtained, based on a differential equations system in the range of continuum mechanics. The novelty of this research is the deduction of the neighborhood structure and rules for a probabilistic CA from these differential equations that describe the evolution of the tumor growth. In addition, the influence of the stresses on tumor growth is captured by the CA.
Uploads
Papers by Aura Conci